Cardiac arrest and CPR are followed by coma and early death in 30-60 percent of cases; and in permanent brain damage in about 20 percent of survivors. Recent animal studies demonstrate a brain damage ameliorating effect of thiobarbiturate loading immediately post-arrest. Recent uncontrolled trials on 90 patients in 4 institutions demonstrate the feasibility of this therapy and suggest its brain damage ameliorating effect. This controlled study is a randomized trial in 12 hospitals of 9 countries, coordinated from Pittsburgh. OBJECTIVES: (1) Determine whether large-dose thiobarbiturate, loading after severe global ischemic-anoxic insults will significantly increase the proportion of CPR survivors with good cerebral performance. (2) Identify possible complications caused by the treatment. (3) Determine one year mortality and morbidity. (4) Set up a collaborative study mechanism for the rapid evaluation of promising brain resuscitation measures prior to their widespread clinical use. Unresponsive patients after CPR are selected and assigned at random to control or treatment groups. Both groups receive the same intensive care. The treatment group, in addition, receives thiopental up to 30 mg/kg once, as early as possible and as rapidly as the circulation can tolerate with blood pressure support. We obtain and analyze at least 100 detailed case reports per year. Sample size estimate in 300 cases maximum. This is based on considering as clinically relevant, an increase in the percent of patients with good cerebral performance categories (CPC 1 and 2 on a 1-5 scale) from 20 percent of patients without, to at least 40 percent with barbiturate. Additional analyses include differences between groups in individual CPCs at certain time intervals up to 12 months; mortality due to cerebral versus non-cerebral causes; and early predictive criteria. Comparisons will be made for control and treatment groups, including stratification into severe and moderate insults. The study is planned for three years (1979-1982) with a fourth year for data analysis.